Finalized FDA Requirements for Standardized Data Max Kanevsky, Sergiy Sirichenko NJ CDISC User Group meeting January 28, 2015

Disclaimer The views and opinions presented here represent those of the speaker and should not be considered to represent advice or guidance on behalf of the Food and Drug Administration.

Abbreviations DSC – FDA Data Standards Catalog SDTCG – FDA Study Data Technical Conformance Guide IG – Implementation Guide (e.g., SDTM IG) SDRG – Study Data Reviewers Guide ADRG – Analysis Data Reviewers Guide

Topics Binding Guidance Timelines SDTCG Requirements

FDA Guidance documents require Standardized Data 2014-12-17 2 Guidance Providing Regulatory Submissions in Electronic Format – Submissions Under Section 745(a) of the Federal Food, Drug, and Cosmetic Act Providing Regulatory Submissions in Electronic Format – Standardized Study Data

4 technical specification documents Data Standards Catalog Study Data Technical Conformance Guide: Technical Specifications Document FDA-specific SEND Validation Rules FDA-specific SDTM Validation Rules http://www.fda.gov/forindustry/datastandards/studydatastandards/default.htm

Timelines Required for NDA, ANDA, BLA, and IND Tied to release date of final guidance Applied to start date of study +2 years for NDA, ANDA and BLA +3 years for IND DSC defines FDA-supported standards and applicable dates http://www.fda.gov/downloads/ForIndustry/DataStandards/StudyDataStandards/UCM340684.xlsx

Formal interpretation “We can wait for January, 2017 before starting implementation of standardized data for new studies”

Real interpretation FDA gives the industry 2-year grace period to implement standards and perfect their process. New guidance documents give FDA the power to reject non-standardized or not compliant data.

Planning and Communication Sponsor/FDA meetings Pre-IND EOP II Type B/C pNDA Sponsor can submit sample data before pNDA

Supplemental documents Study Data Standardization Plan Pre-IND Under development by PhUSE Study Data Reviewer’s Guide Analysis Data Reviewer’s Guide See PhUSE for SDRG and ADRG templates

SDTCG #1.2 Purpose “This Guide provides technical recommendations to sponsors for the submission of … study data and related information in a standardized electronic format ... The Guide is intended to complement and promote interactions between sponsors and FDA review divisions. However, it is not intended to replace the need for sponsors to communicate directly with review divisions regarding implementation approaches or issues relating to data standards…”

#3 Exchange Format XML PDF Documents Supported version in Standard Catalog File Transport Format SAS XPORT v5, not SAS CPORT One dataset per file Dataset name the same as XPT file Common issue in non standardized data like PK, PG

Dataset Size Dataset > 1 GB should be split in smaller files Submit both non-split and split datasets “split” folder (see section #7) Dataset Column Length Maximum length of variable used Variable and Dataset Descriptor Length SAS limitations: Variable Name - 8 Chars Variable and Dataset Label - 40 Chars

Special Characters: Vars and Datasets ASCII text codes only Variable and Dataset Names No punctuation, dashes, spaces or other non-alphanumeric symbols Variable and Dataset Labels May include punctuation characters No special characters like Unbalanced apostrophe, quotation marks, parentheses, braces, brackets “<“ and “>” signs “Parkinson’s”

#4 Study Data Submission Format #4.1 CDISC Currently ADaM specifications for SEND have not been developed Standards Catalog provides a listing of supported data standards and versions Analysis files are critical When IGs do not provide specific instructions on certain study data, Sponsor should discuss with FDA

#4.1.1 SDTM “It is recommended that sponsors implement the SDTM standard for representation of clinical trial tabulation data prior to the conduct of the study. The use of case report forms that incorporate SDTM standard data elements (e.g., Clinical Data Acquisition Standards Harmonization (CDASH)) allows for a simplified process for the creation of SDTM domains.”

“If there is uncertainty regarding implementation, the sponsor should discuss application-specific questions with the review division and general standards implementation questions with the specific center resources identified elsewhere in this Guide (See section 1.2). When data imputation is utilized, sponsors should submit imputed data in an analysis dataset, and the relevant supporting documentation (e.g., ADRG, define.xml) explaining the imputation methods.” “Except for variables that are defined in the SDTMIG as being coded, no numerically coded variables should typically be submitted as part of the SDTM datasets”

SUBJID ID of the entity (i.e., person) in trial If the same subject is screened more than once, then SUBJID should be different USUBJID Unique across the entire application The same USUBJID across all datasets Common issue – PK, PG, SDTM/ADaM, ISS No leading or trailing spaces No inconsistency in usage “0”(S01-001 vs. S1-1) Improper implementation may results in request for Sponsors to re-submit their data

Adjudication Data There are no existing standard or best practice Advised that Sponsors should discuss their approach with review division include details in SDRG presence implementation approach location

#4.1.1.3 SDTM Domain Specifications SUPPQUAL Should be used only for key data which do not fit SDTM domains Examples common issues: SUPPQUALs keep all EDC variables 300+ SUPPQUAL variables Discuss with review division, document in SDRG

DM (Demographics) Single records per subject ARM is blank for Screen Failures “SCRNFAIL” -> “” ACTARM is blank for Not Treated subjects “NOTTRT” -> “” DS (Disposition) EPOCH should be populated for all records DEATH should be the last subject record

SE (Subject Elements) Should be included AE (Adverse Events) Treatment Emergent Flag is expected All AEs from CRF should be included AE Seriousness Criteria should be provided. This info is critical Custom domains Confirm that there are no standard domains Recent versions of IG Provide details in SDRG

LB (Lab Test Results) Large size is common issue Split to < 1GB files according to LBCAT and LBSCAT if needed Trial Design (TS, TA, TE, TV, TI) Should be included

#4.1.2 Analysis Data Model “Generally, ADaM analysis datasets facilitate FDA review. However, it does not always provide data structured in a way that supports all of the analyses that should be submitted for review. For example, ADaM does not support simultaneous analysis of multiple dependent variables or correlation analysis across several response variables. Therefore, sponsors should, as needed, supplement their ADaM datasets after discussions with the specific review division.” Traceability and detailed documentation are expected

Key Efficacy and Safety Variables Efficacy datasets are expected Documentation Timing Variables In addition to protocol-schedules visit variable, at least two additional timing variables are expected AVISIT, AVISITN Core Variables Should be listed after key variables Dates Numeric In addition to ISO8601 Imputation, Flag, documentation

Labels Should be unique and different from SDTM “Adverse Events” is not correct label for ADAE Software Programs Should be provided Used to create ADaM datasets and TFLs ASCII text or PDF files File names with reference to software “adae.sas.txt” Sufficient documentation

ADSL (Analysis Data Subject Level) Required Study specific important baseline subject characteristics and covariates presented in protocol Imputed data When data imputation is utilized, it should be submitted in analysis datasets Detailed relevant supporting documentation Define.xml ADRG Complicated algorithms, etc.

#4.1.3.3 SEND Similar to SDTM MIORRES -> MISTRES modifiers MA domain expect to use VISITDY tumor.xpt is expected for oncology studies

#4.1.4 General Considerations “For the purposes of SDTM and SEND submissions, all Required, Expected, and Permissible variables that were collected, plus any variables that are needed to compute derivations, should be submitted. SDTM datasets should not contain imputed data. FDA recognizes that SDTM contains certain operationally derived variables that have standard derivations across all studies (e.g., --STDY, EPOCH). If the data needed to derive these variables are missing, then these variables cannot be derived and the values should be null.”

Examples of FDA expected variables Baseline flags LB, VS, EG, PC, MB If data were collected or can be derived EPOCH Study Days When --DTC, --STDTC, --ENDTC collected, then populate --DY, --STDY, --ENDY

#4.1.4.5 Data Definition File “define.xml” “A properly functioning define.xml file is an important part of the submission of electronic study datasets.” define.pdf is also expected for define.xml v1.0 Send test files to FDA eData team prior submission An insufficiently documented define file is a common deficiency Stylesheet files for define.xml are required

#4.1.4.6 Annotated Case Report Form New name “acrf.pdf” Mapping of each field on CRF to dataset variable aCRF should include variable names and coding for each CRF item If some data are recorded on CRF but not submitted annotate with “NOT SUBMITTED” text explain in SDRG

#5 Therapeutic Area Standards “This section is reserved for future comments, recommendations, and preferences on therapeutic area data standards.”

#6 Terminology “Common dictionaries should be used across all clinical studies and throughout the submission for each of the following:” AE, CM, PR, MH, indications and study drug names See Standards Catalog for recommended usage of terminology Conformance with Standard Terminology is required Common issues: Misspelling Not following upper/lower case Use of hyphens

Use of Control Terminology “use the most current version of an FDA-supported terminology available at the time of coding.” Different studies may use different versions Impact of usage of old versions should be described in SDRG and Standardization Plan Pooled analysis (e.g., ISS) must use a single version of terminology Maintenance of Controlled Terminology “good terminology management practice” Creation of custom terms is discourage Consistency throughout the application Standardization Plan, SDRG

Adverse Events MedDRA Exact spelling and capitalization Single version for ISS Medications FDA Unique Ingredient Identifier (UNII) TS domain, TSPARMCD=TRT, COMPTRT, CURTRT, … WHO Drug Dictionary CMDECOD – generic name CMCLAS – class or ATC level 4 ATC codes in SUPPCM

Pharmacologic Class National Drug File – Reference Terminology (NDF-RT) TS domain, TSPARMCD=PCLAS Indication SNOMED CT TS domain, TSPARMCD=INDIC, TDIGRP Harmonization with Structured Product Labeling (SPL)

#7 Electronic Submission Format eCTD define.xml and supportive stylesheets in the same folder as datasets No empty folders New “misc” folder is introduced instead of “listings” For need of additional folders consult with FDA

#8 Data Validation and Traceability “data validation is a process that attempts to ensure that submitted data are both compliant and useful. Compliant means the data conform to the applicable and required data standards. Useful means that the data support the intended use (i.e., regulatory review and analysis). “ Study Data Validation Conformance validation and Quality checks Links to FDA rules on Standards Web page Sponsors should fix issues and explain in SDRG why certain errors could not be corrected

Study Data Traceability Important component of regulatory review Relationship between Analysis results Analysis datasets Tabulation datasets Source data Standards are helpful (CDASH) Standardized data will be required Traceability issues with Legacy data conversion

Questions Max Kanevsky mkanevsky@pinnacle21.net Sergiy Sirichenko ssirichenko@pinnacle21.net